Flow-Acoustic Correlation of Turbulent Flow in Pipelines Using Deep Learning

Date
2017
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This thesis considers the development of a proposed pipeline monitoring approach based on acoustic measurements of a pipe. Relationship between the acoustics generated by a turbulent pipeline and the flowrate is examined to understand the physical behaviour of the phenomenon and verify assumptions. A framework is developed to extract features from the flow acoustics in offline and real-time settings for continuous monitoring. To ensure these features are suitable for modelling a flow-acoustic correlation, deep learning and empirical models are compared from experimental measurements of turbulent pipe flows. For deeper insight to turbulent flows, the spatio-temporal dynamics of the flow and acoustics are presented. Empirical dynamic models are shown to predict the dynamics of turbulent flow. The results show experimental evidence of ordered structures in turbulence captured in the acoustics. By isolating these structures, the turbulent motion can be predicted.
Description
Keywords
Artificial Intelligence, Engineering--Electronics and Electrical
Citation
Ma, K. (2017). Flow-Acoustic Correlation of Turbulent Flow in Pipelines Using Deep Learning (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/26190